Register now After registration you will be able to apply for this opportunity online.
This opportunity is not published. No applications will be accepted.
Paramotor Controller Design via Bayesian Optimization
We apply constrained Bayesian optimization algorithms to a RC paramotor to automatically design a suitable controller which improves the efficiency and performance of autonomous flight.
Previous work has shown that a RC paramotor could autonomously track a reference trajectory using a relatively simple control scheme. The controllers that were tested were based on a model that does not capture the paramotor behavior sufficiently. We want to move to an iterative data-driven controller design method. First, we design a suitable control scheme. Then, we use constrained Bayesian Optimization to identify the desired controller parameters. Ideally, the paramotor will follow a reference trajectory, respect maximum pitch angle constraints and maximize its energetic efficiency. The research will focus on including the knowledge about wind speed and direction in a parallelized optimization framework.
The project will have to take place in a practical form. This will require the candidate to be physically available for testing and data collection on the paramotor.
The candidates are required to have good knowledge of Matlab. Hands-on problem solving skills are necessary. Previous experience with micro-controllers is an advantage.
Previous work has shown that a RC paramotor could autonomously track a reference trajectory using a relatively simple control scheme. The controllers that were tested were based on a model that does not capture the paramotor behavior sufficiently. We want to move to an iterative data-driven controller design method. First, we design a suitable control scheme. Then, we use constrained Bayesian Optimization to identify the desired controller parameters. Ideally, the paramotor will follow a reference trajectory, respect maximum pitch angle constraints and maximize its energetic efficiency. The research will focus on including the knowledge about wind speed and direction in a parallelized optimization framework.
The project will have to take place in a practical form. This will require the candidate to be physically available for testing and data collection on the paramotor.
The candidates are required to have good knowledge of Matlab. Hands-on problem solving skills are necessary. Previous experience with micro-controllers is an advantage.
- Apply a data-driven controller design scheme to a RC paramotor to improve it's stability and performance
- Include wind information in the scheme to allow for efficient autonomous flight in moving air
- Apply a data-driven controller design scheme to a RC paramotor to improve it's stability and performance - Include wind information in the scheme to allow for efficient autonomous flight in moving air
Please send your CV and transcript in PDF format to guidetti@inspire.ethz.ch
Please send your CV and transcript in PDF format to guidetti@inspire.ethz.ch